The role of insomnia in the association between posttraumatic stress disorder and hypertension

Abstract

Objective:

Posttraumatic stress disorder (PTSD) is associated with incident hypertension. Although this relationship is poorly understood, PTSD is also associated with insomnia symptoms, which increases the risk for hypertension. Whether insomnia contributes to PTSD-associated risk for hypertension is unknown.

Methods:

We examined self-report survey and electronic health record data from 1109 participants in the Women Veterans Cohort Study (mean age: 43.8 ± 10.9 years; 52% women, 81% White) to assess the cross-sectional associations between PTSD symptom severity, recent symptoms of insomnia, and hypertension, defined as self-reported treatment for high blood pressure in the last year. Structural equation modeling was used to examine whether insomnia symptoms mediate the association between PTSD and hypertension.

Results:

PTSD symptom severity was associated with hypertension (r = 0.09, P < 0.001). PTSD symptom severity and hypertension were each associated with the insomnia symptoms difficulty falling asleep, difficulty staying asleep, and worry/distress about sleep problems (PTSD: rs = 0.58–0.62, P < 0.001; hypertension: rs = 0.07–0.10, P < 0.001). A latent variable derived from those symptoms mediated 9% of the association between PTSD symptom severity and hypertension (P = 0.02).

Conclusion:

In this study of young and middle-aged Veterans, insomnia symptoms mediated the association between PTSD and hypertension. Difficulties falling asleep and maintaining sleep and related distress may be particularly deleterious for cardiovascular health in Veterans. Longitudinal data is required to further investigate the associations between PTSD, insomnia, and hypertension.

INTRODUCTION

More than half of United States military Veterans will develop hypertension in their lifetime [1]. Hypertension is the most common chronic condition in this group [2], and has an earlier age of onset in Veterans compared with the general population [3,4]. Posttraumatic stress disorder (PTSD) is also common (23%) among younger Veterans [5], and is associated with a greater risk of hypertension [6,7], and a higher prevalence of incident cardiovascular disease [7].

Although the pathophysiology linking PTSD to incident hypertension has not been established, disruption of sleep, and potentially insomnia, may be involved. Insomnia is a clinical syndrome characterized by recurrent episodes of difficulty initiating or maintaining sleep [8], and is associated with a greater risk of incident hypertension [9,10]. Overall, an estimated 20% of military service members endorse symptoms of insomnia [11] with the highest rates observed in younger Veterans who served in the recent Iraq and Afghanistan conflicts [e.g. Operation Enduring Freedom (OEF), Operation Iraqi Freedom (OIF), and Operation New Dawn (OND)] [12]. The prevalence of insomnia is even higher among those with PTSD [9]. It is, therefore, conceivable that insomnia may in part contribute to the association between PTSD and cardiovascular disease.

Multiple pathways may link PTSD and insomnia to incident hypertension risk. Disturbed sleep and nightmares are among the diagnostic criteria for PTSD [13], and thus, these insomnia symptoms of PTSD may in particular, carry the PTSD-associated hypertension risk. Bidirectional associations between PTSD and insomnia are also possible [14,15], but have not been comprehensively explored in relation to cardiovascular risk in Veterans. Finally, not all individuals with PTSD report insomnia symptoms [14,15]. Thus, there could be phenotypes of PTSD in which some Veterans have poor sleep and others do not, and those with poor sleep are more vulnerable to hypertension. Given the increasing burden and potential socioeconomic consequences of early-onset hypertension, further research is needed to clarify the role of sleep in this association to inform future interventions in this at-risk group.

In this study, we used cross-sectional self-report data from a sample of Veterans who participated in OEF/OIF/OND to evaluate: the associations between recent symptoms of insomnia, PTSD symptom severity, and recent treatment for high blood pressure (i.e. hypertension), and whether insomnia symptoms mediated the association between PTSD and hypertension. We hypothesized that specific insomnia symptoms, including problems falling and staying sleep, would mediate this association. Due to known sex differences in insomnia and hypertension [16,17], the mediating role of insomnia symptoms was also explored separately for men and women Veterans. Finally, we tested the direction of mediation by considering PTSD and insomnia as both predictor and mediator variables.

METHODS

Study design and population

The Women Veterans Cohort Study (WVCS) is a multisite prospective study designed to identify sex-associated disparities in health outcomes and healthcare utilization among young OEF/OIF/OND Veterans discharged from military service since 1 October 2001 and who enrolled in Veterans Health Administration (VHA) healthcare [18]. The study cohort was derived from a roster of OEF/OIF/OND Veterans maintained by the VHA and the Defense Manpower Data Center – Contingency Tracking System Deployment File. As depicted in the study flowchart (Fig. 1), 1109 from the parent cohort completed a follow-up survey (52% women). Survey participants received $20 in compensation. The Institutional Review Boards of participating VHAs approved all study procedures.

FIGURE 1.

A flowchart of the Women Veterans Cohort Study (WVCS) electronic health record (EHR) sample and survey recruitment. A random sample of men and women enrolled in the parent EHR cohort were recruited from Veterans Health Administration medical centers across the United States (West Haven, Connecticut; Durham, North Carolina; Indianapolis, Indiana; and Los Angeles, California) to participate in the survey. To increase participation by women Veterans, women were over-recruited at a ratio of 1.5 : 1. Eligible participants were mailed an invitation to participate, along with consent documents, and a paper version of the baseline survey, between 2013 and 2015. Mailings were resent up to three times to those who did not reply to the initial invitation. Follow-up surveys were mailed annually for the subsequent 2 years (data not reported herein). Of those Veterans who were invited to participate in the survey cohort (n = 8465), 1109 (52% women) completed the baseline survey, a response rate of 13%.

Survey measures and variables

Demographic and clinical characteristics, health behaviors, and military service

Self-reported survey data included participant age at the time of the survey, sex, race, ethnicity, marital status, education, and employment. General health, smoking status, BMI and the number of deployments during military service were also reported.

Veterans RAND 12-Item Health Survey

The Veterans RAND 12-item health survey (VR-12) [19] is a measure of general health status including physical and social functioning, role limitations, pain, and mental health. The VR-12 provides an aggregate score of mental and physical health.

PTSD checklist – Civilian version

The PTSD checklist – Civilian version (PCL-C) [20] is a 17-item questionnaire used to assess the presence and severity of PTSD symptoms ‘that Veterans sometimes have in response to stressful life experiences’. Participants indicate how much they have been afflicted with a symptom in the last month, based on criteria from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV). Endorsed symptoms are totaled. For Veterans treated in VHA primary care a total of at least 45 distinguishes a PTSD diagnosis [21].

Insomnia Severity Index

The Insomnia Severity Index (ISI) [22] is a seven-item measure that is designed to assess recent symptoms of insomnia (i.e. over the last 2 weeks). A total of at least 15 indicates clinical insomnia. A total score, along with responses to specific items (difficulty falling or staying asleep, waking too early, satisfaction with sleep patterns, sleep interfering with quality of life, and worry/distress related to sleep) were examined in these analyses. The ISI is a well-validated measure of insomnia according to the International Classification of Sleep Disorders and has been positively associated with both sleep diary and polysomnographic variables [22].

Hypertension

Participants replied to the question, ‘In the past 12 months, have you received medical treatment for high blood pressure?’ Confirmatory responses were classified as indicative of a hypertension diagnosis.

Data obtained from electronic health records

International Classification of Diseases (ICD) 9th and 10th Revision codes were used to identify a diagnosis of hypertension in the EHR. Comparisons of subjective reports with clinical diagnoses of hypertension have shown good to excellent concordance between sources [23,24]. In this sample, about 18% of the population had a diagnosis of hypertension and there was good concordance between hypertension diagnoses and subjective endorsement of recent treatment for high blood pressure (κ = 0.60, P < 0.001), with a sensitivity of 89% and specificity of 76%.

As obstructive sleep apnea (OSA) is another sleep disorder that is associated with hypertension [25], a diagnosis of OSA was also ascertained from the EHR.

Statistical analyses

Demographic and clinical characteristics are presented in the overall sample and according to whether an individual met the diagnostic threshold for insomnia on the ISI. Categorical variables are presented as frequencies and percentages, and continuous variables are presented as mean ± standard deviation.

At least 100–200 cases are recommended to test mediation [26]. A power analysis was conducted with G*Power (Version 3.1.9.2; Bonn University, Bonn, Germany). With a small effect size ƒ²(0.02), α = 0.05, power = 0.80 and two predictors, 485 participants were required for these analyses. This sample exceeded both criteria.

Continuous variables generated from the PCL-C and ISI were evaluated for skewness and an absolute value less than 3 was considered univariate normal [26]. PTSD and insomnia symptom severity were each included as continuous variables in mediation analyses. Potential confounding variables were chosen a priori based on previous investigations of PTSD and hypertension. These included self-reported number of deployments [27,28], a diagnosis of OSA [29,30], and sex [31,32]. Other factors may confer cardiovascular risk (e.g. poor diet, insufficient exercise, stress) but accounting for those factors is beyond the scope of this dataset. Additional covariates were added to the model if there were significant differences between Veterans meeting self-report diagnostic criteria for insomnia and those without insomnia.

Mediation was tested using the approach of Baron and Kenny [33]. In the first step, we determined if PTSD symptom severity was significantly associated with hypertension. The second step required demonstrating that PTSD symptom severity was significantly associated with individual symptoms of insomnia. During the third and fourth steps, we used multiple regression to assess, if individual insomnia symptoms were significantly associated with hypertension when controlling for PTSD and if the association from PTSD symptom severity to hypertension, controlling for insomnia, was smaller than the association from PTSD to hypertension alone.

Structural equation modeling and latent factor analysis with full information maximum likelihood were used to estimate model parameters while accounting for data missing at random (≥5%). Models were estimated using boot-strapping with 10 000 iterations and 95% confidence intervals to determine the significance of the effects. A latent variable was generated based on the symptoms of insomnia that were significantly associated with both PTSD symptom severity and hypertension. A confirmatory factor analysis was first performed to verify adequate measurement of the latent insomnia variable. An exploratory factor analysis was then used to assess if this latent variable mediated the association between PTSD and hypertension. Model fit was assessed with several indices, per published recommendations [34], including the chi-square test (<0.05), comparative fit index (CFI >0.90), standardized root mean square residual (SRMR <0.08), and root mean square error of approximation (RMSEA <0.08).

Missing data

Approximately 4.6% of the data were missing. For variables with less than 5% missing data, a regression-based multiple imputation technique was used to generate five imputed data sets, and the results were averaged. On the basis of t-tests, there were no significant differences between the original and imputed variables (all Ps > 0.05).

Descriptive and bivariate analyses were performed with Stata (v12.1; Stata Corporation, College Station, Texas, USA) and path analyses were conducted using Mplus software (Muthén & Muthén, Los Angeles, California, USA). An α level of P < 0.05 was selected to determine statistical significance.

RESULTS

Characteristics of the study sample are depicted in Table 1. The mean age was 43.8 years ± 10.9 and over half were women (51.8%). A high proportion were White (81.3%), non-Hispanic (88.5%), and were married or in a relationship (55.4%). A majority reported greater than a high school education (79.8%) and being employed (60.6%). The sample also reported an average of two combat deployments during military service. A small percentage were current smokers (13.5%) and the entire group had an average BMI of 28.9, falling within the overweight range. On the PCL-C, a third of the sample scored in a range consistent with a diagnosis of PTSD (n = 360, 32.5%), which was within the expected range for this population (16–39%) [21]. Similarly, one third of the sample screened positive for insomnia on the ISI (n = 373, 33.6%), of which 45.3% met criteria for PTSD (n = 169). About 19% of Veterans endorsed recent treatment for high blood pressure and were thus classified as having hypertension (n = 213). Approximately 24% (n = 52) of Veterans with self-reported hypertension screened positive for both insomnia and PTSD.

TABLE 1.

Demographic characteristics, lifestyle factors, self-reported health, and clinical factors according to insomnia diagnosis

	Total sample (N=1109)	Insomnia (n=373)	No insomnia (n=736)	P-value
Demographics
Age (years)a	43.79±10.89	42.72±10.67	44.33±10.96	0.020
	Range: 23–71
Female (%)	575 (51.8)	195 (52.3)	380 (51.6)	0.880
Race (%)
White	902 (81.3)	285 (76.4)	617(83.8)	0.060
African American	107 (9.6)	54 (14.5)	53 (7.2)
Other	100 (9.0)	34 (9.1)	66 (9.0)
Hispanic (%)	127 (11.5)	49 (13.1)	78(10.6)	0.312
Marital status (%)				0.125
Married/in a relationship	614 (55.4)	194 (52.0)	420 (57.1)
Single	187 (16.9)	98 (26.3)	89 (12.1)
Divorced/other	308 (27.8)	81 (21.7)	227 (30.8)
Education (%)				0.049
High school or less	225 (20.2)	86 (23.1)	139 (18.9)
Some college	260 (23.4)	90 (24.1)	170 (23.1)
College degree or higher	624 (56.3)	197 (52.8)	427 (58.0)
Employment (%)
Employed	672 (60.6)	194 (52.0)	478 (64.9)	<0.001
Student/retired/unable to work	364 (14.2)	142 (38.1)	216 (29.3)
Unemployed	73 (14.6)	37 (9.9)	42 (5.7)
Military service – times deployeda	2.11±1.60	2.14±1.56	2.10±1.60	0.710
	Range: 0–15
Lifestyle factors
Smoking (%)				0.115
Never	591 (53.3)	189 (50.7)	402 (54.6)
Former	367 (33.1)	127 (34.0)	240 (32.6)
Current	151 (13.5)	57 (15.3)	94 (12.8)
BMIa	28.91±5.56	29.27±5.87	28.75±5.35	0.141
	Range: 13–56
Self-reported general health
VR-12	53.89±24.39	48.51±26.64	56.62±22.71	<0.001
	Range: 0–100
Self-reported mental health symptoms
PCL-C Scorea	38.25±17.59	45.97±21.21	35.19±14.00	<0.001
	Range: 17–85
PCL-C diagnosis of PTSD (%)	360 (32.5)	169 (45.3)	191 (26.0)	<0.001
ISI scorea	10.75±5.49	15.15±5.05	8.28±3.45	<0.001
	Range: 1–24
ISI diagnosis of insomnia {%)	373 (33.6)	-	-	-
Clinical factors
Obstructive sleep apnea (%)	23 (2.1)	13 (3.5)	10 (1.4)	0.042
Treatment for high blood pressure (%)	213 (19.2)	89 (23.9)	124 (16.8)	0.206
EHR diagnosis of hypertension (%)	197 (17.8)	71 (19.0)	126 (17.1)	0.501

N = 1109. Bold values are significant. EHR, electronic health record; ISI, Insomnia Severity Index; PCL-C, PTSD Checklist – Civilian Version.; VR-12, Veterans RAND 12-Item Health Survey.

^aData is presented as mean ± SD.

When comparing Veterans based on self-reported insomnia, those meeting diagnostic criteria were significantly younger, more educated, and more likely to be employed than those without insomnia (Ps <0.001–0.049), but there were no significant group differences in sex, race, ethnicity, or marital status. PTSD and OSA were significantly more common among those with insomnia (Ps < 0.001), but the groups did not differ in smoking status, BMI, hypertension, or number of deployments (Table 1). To avoid multicollinearity between employment and education, only the age and employment variables were added as covariates to the structural equation models.

PTSD symptom severity was associated with hypertension (r = 0.09, P < 0.01), although the magnitude of the relationship was small. PTSD was strongly correlated with all individual ISI items (Ps < 0.001; Table 2), and there were small-to-moderate correlations between the ISI items and hypertension. Symptoms of difficulty falling asleep and staying asleep, waking too early, and worry/distress related to sleep were each significantly associated with hypertension when controlling for PTSD (Ps = 0.02–0.001). Lastly, controlling for the insomnia symptoms, the association of PTSD to hypertension, was reduced (R² = 0.03 vs. R² = 0.04; Ps < 0.001).

TABLE 2.

Zero-order correlations between PTSD, symptoms of insomnia, and hypertension

Variable	1	2
1. PCL-C score	1
2. Hypertension, 1=endorsed	0.090**	1
3. Difficulty falling asleep	0.619**	0.074*
4. Difficulty staying asleep	0.581**	0.120**
5. Waking too early	0.476**	0.103**
6. Dissatisfaction with sleep	0.521**	0.054
7. Sleep-related quality of life impairment	0.544**	0.038
8. Sleep-related worry/distress	0.608**	0.067*
9. Sleep interferes with daily functioning	0.625**	0.058

N = 1109. PCL-C, PTSD Checklist – Civilian Version.

^*P < 0.05.

^**P < 0.01.

Confirmatory factor analysis was conducted with the four significant insomnia symptoms to determine if a latent insomnia variable would approximate those symptoms. Fit indices for the latent insomnia variable ranged from excellent to poor [χ²(2) = 56.58, P<0.001; CFI = 0.98; RMSEA = 0.16; 90% CI (0.06–0.26), SRMR = 0.02], indicating the variable was not well represented by the four items. Waking too early had the lowest factor loading (β = 0.72, P<0.001), and thus was trimmed. The latent insomnia variable was well represented by these three symptoms.

Fit indices for the exploratory mediation model indicated excellent statistical fit [χ²(23) = 137.97, P<0.001; CFI = 0.95; RMSEA = 0.07, 90% CI (0.06–0.08); SRMR = 0.05]. Standardized factor loadings of the three insomnia symptoms on the latent variable of insomnia symptoms were each statistically significant [βs = 0.79–0.83, Ps<0.001], indicating that the latent insomnia variable was again well approximated within the full model.

Standardized direct and indirect associations were examined next (Fig. 2). R² was 52.9%, representing the percentage of PTSD symptom severity accounted by the model. The direct association of PTSD on hypertension was not significant [β = 0.02; 95% CI (−0.08 to 0.12); P = 0.66], but the indirect association was significant, with insomnia symptoms accounting for 9% of the variance in hypertension [β = 0.09; 95% CI (0.02–0.17); P = 0.02].

FIGURE 2.

Results from the mediation model. Observed variables are represented by rectangles and a circle represents the latent variable. The predictor, mediator and outcome variables and their associated direct and indirect associations are depicted. Symptoms of insomnia mediated the pathway between PTSD (i.e. PCL-C symptoms) and hypertension, accounting for 9% of the association. PCL-C, PTSD Checklist – Civilian Version; PTSD, posttraumatic stress disorder.

Exploratory factor analyses were also run separately for men and women. For both groups, the direct association between PTSD symptoms and hypertension was again not statistically significant [men: β = 0.04, 95% CI (−0.05 to 0.12); P = 0.44; women: β = 0.10, 95% CI (−0.07 to 0.16); P = 0.44], and insomnia symptoms mediated the association between PTSD symptoms and hypertension [men: β = 0.08, 95% CI (0.01–0.15); P = 0.03; women: β = 0.10, 95% CI (0.01–0.18); P = 0.03].

Given that these data are cross-sectional, the PTSD and insomnia variables were exchanged as predictor and mediator, and the model was rerun with the entire sample. Fit indices worsened [χ²(28) = 180.11, P<0.001; CFI = 0.93; RMSEA = 0.71, 90% CI (0.06–0.08); SRMR = 0.06]. Although the direct association of insomnia symptoms and hypertension remained statistically significant [β = 0.13, 95% CI (0.03–0.21); P = 0.01], PTSD symptoms did not mediate the association between insomnia symptoms and hypertension [β = 0.01, 95% CI (−0.06 to 0.07); P = 0.70].

DISCUSSION

In this cross-sectional study of younger Veterans there are two main findings: PTSD and insomnia were both associated with hypertension, defined as receiving treatment for high blood pressure, and insomnia symptoms mediated the association between PTSD and hypertension. Importantly, mediation remained when examining men and women Veterans separately. Thus, for young Veterans, difficulty falling and staying asleep and worry about sleep could be important predisposing factors for hypertension whereas other symptoms of insomnia, such as problems of waking too early, nonrestorative sleep, and interference with daily functioning may confer less risk. Individual symptoms of insomnia may be more efficacious targets for intervention to reduce Veterans risk of developing hypertension.

The association between PTSD and insomnia is well documented [14,15], and indeed, symptoms of insomnia – nightmares, difficulty falling asleep – are among PTSD diagnostic criteria [13]. According to the 3-P (precipitating, predisposing, and perpetuating) Model of insomnia [35], certain individuals may be predisposed to insomnia based on characteristics including rumination or hyperarousal, each of which is seen in PTSD [13]. Stressful life events may contribute to the onset or progression of insomnia symptoms. Additional factors, such as poor sleep hygiene or ineffective coping skills may increase psychological and physiological arousal, exacerbating insomnia. This model was developed with civilian populations but may be especially relevant to young Veterans who were exposed to frequent, intense stressors during their recent service in Iraq and Afghanistan [36]. Consider that individuals with PTSD and insomnia are prone to rumination (e.g. worry or distress about sleep, as observed in this study), and to generalized and sleep-related anxiety [13,37]. Experiences during military service, including the heightened stress characteristic of deployment and combat zones, worsen symptoms of insomnia [27]. Perseverating about physical symptoms (i.e. insomnia or high blood pressure) may further promote rumination, anxiety, and hyperarousal in a self-perpetuating cycle that could increase an individual’s hypertension risk. In turn, individuals with PTSD and insomnia exhibit increased sympathetic nervous system activity and a greater risk of incident hypertension [38,39]. The symptoms of insomnia that are associated with hypertension are not limited to the characteristics of sleep assessed on the PCL-C (i.e. repeated disturbing dreams, trouble falling or staying asleep), thus highlighting a distinct role for insomnia.

Although these data were cross-sectional, we speculate that PTSD may be an early precursor for insomnia, which in turn, can lead to incident hypertension. When symptoms of PTSD and insomnia were exchanged in the model, only the direct association from insomnia to hypertension remained significant. Thus, the insomnia symptoms reported by this sample appeared to represent a more robust, separate factor associated with hypertension, instead of secondary manifestations of PTSD. It is also possible that Veterans who endorsed difficulty falling and staying asleep, and worry about sleep, represent a subgroup of patients who are most vulnerable to hypertension, compared to those with different symptoms of insomnia, or an alternate history of chronic or traumatic stress. The direction of these relationships is important to determine, as it has implications for behavioral interventions that target PTSD or insomnia as part of an effort to reduce the associated incident cardiovascular disease risk. Symptoms of insomnia are less likely to remit after treatment for PTSD [40]. Thus, a number of ongoing and completed clinical trials target poor sleep in patients with PTSD and assess consequent effects on PTSD symptom severity [41]. It may behoove researchers to leverage those datasets to explore the effects of sleep treatment on incident hypertension. Alternately, researchers could examine EHR data from patients with and without PTSD and sleep treatment to compare their risk for hypertension.

Clinical implications of findings

From a public health perspective, standardized screening and development of more targeted treatment for insomnia may help decrease the associated risk of cardiovascular disease. According to one investigation, insomnia was the most frequently reported symptom on the PCL-C among OEF/OIF/OND military service members [42]. Veterans may be more willing to endorse sleep problems than PTSD-related symptoms, other symptoms of mental health disorders (e.g. depression), or other risk factors for hypertension (e.g. poor diet, inactivity) [43], creating a golden opportunity for clinicians to assess and treat insomnia and behaviorally target cardiovascular risk. Yet, a recent study of Veterans and VHA primary care providers found that 20–39% of Veterans disclose symptoms of insomnia, but only half of providers regularly record insomnia in the medical record [44]. Administrators could significantly improve cardiovascular prevention and VHA care by ensuring the regular evaluation, documentation, and referral for insomnia and other sleep symptoms. Furthermore, as only 50% of Veterans receive medical services in the VHA [45], intervening on insomnia is equally important in non-VHA healthcare centers where the other 50% of this population may receive treatment.

Limitations

The present investigation has several limitations. First, the cross-sectional design did not allow for broader inferences about causality. Longitudinal studies to evaluate OEF/OIF/OND Veterans predeployment and postdeployment and discharge are required to assess the timing of PTSD and sleep problems, and downwind consequences for blood pressure and hypertension. Second, this sample was limited to OEF/OIF/OND Veterans who had received care in the VHA, which may limit the representativeness of this cohort. Yet, it appears that Veterans in this cohort who are receiving care through the VHA are comparable with those who are not [18]. Third, PTSD, insomnia, and hypertension are each related to psychophysiological hyperarousal [46,47], a factor that was not assessed but one that could play a mediating role for both insomnia and PTSD in relation to hypertension risk. Fourth, self-report data is inherently biased. These analyses were limited to self-report of recent medical treatment for high blood pressure, with ‘medical treatment’ not specifically defined. Yet, the use of self-reported hypertension has been previously validated [48]. Moreover, to understand the relationship between hypertension and recent symptoms of PTSD and insomnia, our analysis prioritized recent treatment for hypertension over a lifetime diagnosis, assuming the former represents recent symptoms. There was acceptable statistical concordance between self-reported hypertension and an EHR diagnosis of hypertension, and we expect that a self-reported diagnosis of hypertension would show a closer association with the EHR diagnosis. Finally, questions on the PCL-C concerning PTSD symptoms limit the time frame to the past month. Thus, the sample may have a different risk of hypertension than Veterans with a lifetime history of PTSD but who were less symptomatic.

In conclusion, in this cross-sectional study, we found that the association between PTSD and hypertension was mediated in part by a specific group of insomnia symptoms. Although these results are preliminary and merit replication with prospective designs in Veteran and non-Veteran samples, the findings support the potential role of insomnia in bridging the health risk incurred because of trauma with cardiovascular health, and as a distinct risk factor for hypertension. These findings also highlight the potential for mitigating PTSD-associated risk for hypertension by targeting comorbidities, such as insomnia. As next steps, regular screening for insomnia and other symptoms of disordered sleep, and determining the value of targeted systemic approaches to sleep treatment in those with PTSD, may better serve the heart health of our Veterans.

Abbreviations:

CFI

comparative fit index

DSM-IV

Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition

EHR

electronic health record

ISI

insomnia Severity Index

OEF/OIF/OND

Operations Enduring Freedom/Iraqi Freedom/New Dawn

OSA

obstructive sleep apnea

PCL-C

PTSD Checklist – Civilian Version

PTSD

posttraumatic stress disorder

RMSEA

root mean square error of approximation

SRMR

standardized root mean square residual

VHA

Veterans Health Administration

VR-12

Veterans RAND 12-Item Health Survey

WVCS

Women Veterans Cohort Study